Numerous analyses of control loops have been carried out in various industries, with most painting the same picture: Only a third of the control loops ran with very good or good performance; the remaining two-thirds were either poorly adjusted or required manual operation either periodically or constantly. In practice, PID parameters are generally determined by heuristic adjustment rules. If data from other systems is available, the user will also frequently rely on these parameter sets. With computer-aided controller optimization, the process is either initiated using a manipulated variable step-change (open control loop) or a setpoint step-change (closed control loop).

Based on the APROL process control system, B&R now offers APROL APC, a ready-to-use solution for the Automation PC 910. As an engineering platform, APROL ensures maximum flexibility with the least amount of cost and effort. Implementing advanced process control has been simplified considerably with this solution.

Because it uses APROL as its platform, APROL APC is suitable for applications from small laboratory facilities to large-scale production plants. An enormous amount of investment protection is the result. Thanks to its flexible scalability, the system can grow to meet new challenges. Whether a couple dozen measurement points or several thousand, APROL always has the perfect system for any application.

 

System topology
APROL APC is delivered on an Automation PC 910 with all of the necessary software already installed. In addition to the engineering software and operatorsoftware, this industrial PC also comprises a high-performance database with an SQL interface running on the extremely stable SUSE Linux Enterprise Server operating system. This system can be installed in a control cabinet without a monitor. Access is provided using a web browser or VNC client from the operator’s workstation. A controller is required to process the APC blocks. Additional controllers can easily be added as needed.

Connectivity guaranteed
POWERLINK provides the communication pathway between the controller and the I/O modules. Other fieldbuses for third-party process control systems such as Modbus TCP, Profibus DP, PROFINET or OPC can also be integrated if required without any problems. Using Automation Studio, programmed controllers can be easily connected using INA/ ANSL communication.

Minimal engineering
All input and output signals are captured using compact X20 I/O modules from B&R’s extensive range. Special-purpose modules such as M-bus, electrical metering modules, vibration measuring modules and modules with HART support are also available. APROL offers users maximum flexibility with minimum engineering expenditure. As the system can be scaled to suit your precise needs, you benefit from the highest possible investment protection.

Flexible integration into existing automation solutions
You can decide whether you want to install Advanced Process Control independently of the existing SCADA systems, process control systems or SPS solutions. APROL APC is designed to be a standalone solution as well as a fully integrated solution. It can be integrated into existing APROL process control systems at any time.

APROL APC – An optimal solution for all industries

• Machine automation
  The best possible optimization of control is at the machine level using APROL APC. If APROL APC is not used for series-produced machines due to cost reasons, the project can still be created with Automation Studio since it also contains the APC blocks.
• Plant automation
  Plant automation refers primarily to medium-sized plants with multiple controllers. Since a lot of value is placed on complete uniformity in such practices, integrating APROL is the optimum solution.
• Process automation
  If you already have an APROL process control system, it is possible to integrate it into the process control system. Existing third-party process control systems can easily be connected via interfaces. Detailed analysis and optimization can take place directly in the installed APROL APC standalone solution.

In addition to an advanced PID/PI²D controller, the comprehensive Process Automation Library (PAL) offers a wide range of powerful control modules.

Control modules for APROL APC
1. PID/PI²D controller

Controller structure
The controller structure is based on the ISA standard. The many different configuration options mean that there are many different controller characteristics. A significant advantage of this structure is that the reference variable behavior can be configured independently of the disturbance variable behavior. This means the following controller structures can be implemented, for example:

The faceplate of the PID/PI²D controller shows all relevant values and supports user interaction.

Highlights

• PID
  Classic PID controller in standard form
• I-PD
  The I component affects control errors; the P and D components affect the controlled variable.
• PI-D
  The P and I components affect control errors; the D component affects the controlled variable.
• Controller with two degrees of freedom

 

Important properties of the PID/PI²D controller
Among other things, the controller function block contains the following important functions:

• Option of standardizing manipulated variable output and reference variable input
• Deadband function (also asymmetrical: support for separate configuration options for positive and negative deviation)
• Possible to configure controller as PID or PID.
  The windup limits, reset time, initial parameter value and option for seamless changeover between operating modes depend on the component that is active (I or I²).
• P, D, and I or I² component can be switched on or off separately.
• Possible to track manipulated variable
• Initialization value of the integrator
• “Anti-windup” (windup limit I component corresponds to the limit I² element); also possible to give the integrator a maximum limit “holdmax” and minimum limit “holdmin”
• A corresponding output is set as soon as the manipulated variable output reaches an upper or lower absolute limit. These outputs can then be used to set the directional limit of the integrator.

Controller operating modes
The following operating modes are built directly into the logic of the function block. Switching between them is bumpless.

• Off
• Manual
• Open
• Closed
• Hold
• Track
• Autotune
• Automatic

Control methods / Controller operating modes
The PI/I²D function block covers the following functions:

• Seamless transfer between operating modes
• Cascade control (tracking)
• Windup protection in the cascade
• Override control
• Manipulated variable as a deciding factor
• Gain scheduling
• Autotuning

2. Autotuning PID control loops

The PID tuner is used to identify the process model automatically. The acquired data is used to determine the control parameters in accordance with the optimum criteria method.

AUTOTUNE mode
The APROL PID controller has a separate AUTOTUNE operating mode, which enables the tuning function block to have direct process access to the PID without additional wiring. Using the function block, the PID control parameters for controlled systems can be easily determined to a sufficiently precise level without needing special technical knowledge.

Step response
With step response (autotuning), the step response of the controlled system is recorded in the open control loop in order to calculate the control parameters. The starting point is the starting value of the manipulated variable, which should hold the controlled variable more or less to the operating point. The manipulated variable allowed during tuning can be specified by setting parameters; in addition, the maximum self-optimization time is defined.

Oscillation tuning
Fatigue testing is an identification procedure in the closed control loop. In this procedure, the system is controlled so that the process variable oscillates periodically around the predefined setpoint. The period duration and amplitude ratio of the manipulated variable and the controlled variable are used to determine the PID control parameters in accordance with Ziegler/Nichols (heuristic process).

 

3. Split range control

With split range control, several actuators monitor a controlled variable. The control input is therefore fed to two different adjustment devices. This means it is possible to control both heating and cooling with just one controller, for example. In addition to the various physical effects, the adjustment devices can also exhibit different techniques. In principle, the split range function consists of two characteristic curves for both actuators. Different gain factors of the final control elements can be balanced out so that the resulting behavior (configuration) for the controller is uniform.

PID control with parameter control that depends on the operating point (gain scheduling)
In many technical processes, it is necessary to adapt to fluctuating parameters, as these demonstrate non-linear behavior due to certain physical, chemical or thermodynamic effects.

It is necessary to adapt to process non-linearities
Non-linearity results in different gain factors or process time constants being effective at various operating points. Various control parameters are deemed to be optimal in this regard.

K, Ti, Td adjusted to the operating point
PID parameter sets that are dependent on working points can be used to control systems with non-linearity. This makes it possible to feed forward the proportional, integral and differential components of a PID controller based on operating points. This simple solution to the described problem is known as “gain scheduling” or “adaptive control”.

If a signal on the reference input is between two configured reference points, then the PID parameters for gain, integration time and derivative time will be interpolated linearly based on the configured reference values. A typical use for gain scheduling is controlling the pH value. The gain factor of the PID controller uses the pH value to steadily adjust piecemeal to the acid-base titration curve.

4. Override control

Multiple controllers affect the same actuator when override control is involved. The controllers are wired so that only one controller is active at a time. A controller is given access to the actuator based on the current process state.

The decision as to which controller is active at a given time can be made either by comparing the two controllers’ control values (min/max selection of the manipulated variable) or by using definable threshold values related to the controlled variable or an external reference variable. In the event of an alternating criterion based on the manipulated variables of both controllers, the controller with the largest or smallest output signal will be given access to the actuator.

Override control is used in compacting systems, for example. These are primarily regulated based on capacity, with pressure also among the monitored values. Above a certain pressure, the performance control is separated from the pressure control, which also has an effect on the motor speed of the entire system.

5. Disturbance variable feed-forward

Disturbance variables measurable / Controlled system dynamics known With disturbance variable compensation, a measurable disturbance variable is used as the starting point. It can be directly compensated using the known dynamics of the controlled system and feed forward control. The following general strategy applies in such cases: Control (open loop) as much as possible (i.e. as much as possible based on knowledge of a process model), regulate (closed loop) as much as necessary (as a result of uncertainty regarding the model and/or disturbances that cannot be measured).

 

6. Cascade control

Primary controller / Secondary controller
Cascade control describes cases where multiple controllers are connected in series. Cascade control utilizes a master controller, which sends the manipulated variable as a setpoint to the downstream slave controllers. This type of connection results in nested control loops that are able to more quickly compensate for any disturbances that occur in the respective control loop. Here, the auxiliary controlled variable is assigned its own controller, which is referred to as the slave controller. The higher-level setpoint tracking controller passes on the reference variable to the subsequent controller. Connecting control loops in series requires that the lower-level control loops feature successively higher dynamics. An example of this is the flow rate control with higher-level temperature regulation or an MPC that acts as a master controller and supplies setpoints for secondary PID control loops. With cascade control, it is important to ensure that the integral component of the setpoint tracking controller is immobilized when limiting the manipulated variables of the subsequent controller. In addition, the adjustment range of the setpoint tracking controller must match the target variable of the subsequent controller. If the subsequent controller is not in automatic operation (with an external setpoint), then the setpoint tracking controller must be tracked so the I component of the setpoint tracking controller does not reach saturation (windup). To ensure bumpless transfer later on, the control value of the setpoint tracking controller must be tracked with the setpoint of the subsequent controller at the time.

 

7. Ratio control

Open controller or closed control loop Ratio controls are used to mix material flows at a defined ratio. There are various options for this since the setpoint for the secondary controller can be formed from the setpoint of the first controller or from its actual value. The former generally leads to the secondary controller performing more smoothly. In both cases, it is an open controller rather than a closed control loop.

Typical uses for ratio control

• All types of mixture control (e.g. gas and air mixtures in combustion processes or acid/alkaline neutralization)

Pilot control through the ratio block is used to set a defined connection between process variables of a plant. The ratio block can also be used for direct pilot control of a second control element (synchronization control). The ratio control is used for all kinds of mixing control application, such as gas-air mixtures for combustion processes or neutralization of acidic and alkaline solutions.

 

9. Smith predictor control for systems with dead time

Considerable dead times not controllable by PID
Dead time can be identified when the controlled variable does not initially react to a manipulated intervention. In processes with long dead time, a standard PI controller must be adjusted very slowly; as a result, corresponding deductions must be made in control quality. This quality can be significantly improved by means of what is known as the Smith predictor.

Model provides additional information
The Smith predictor refers to a controller structure that also draws on information delivered by a model of the system. The Smith predictor controller configuration is suitable for processes with a high degree of dead time. The additional information means the controller can be configured more aggressively, which in turn significantly increases the quality over a conventional controller configuration.

Predefined process models possible
The following process models are available for configuring the dynamic process model: P, PT1, PTDT1, PT2, PTDT2, I, IT1, IPT2DT, PT3, DT1, dead time.

Model-based predictive control is based on an optimization method that cyclically minimizes predicted control errors. Predicting control errors is backed up by the use of a model. To do so, pulse and/or step responses are saved in the control function block. The calculation is based on the assumption that the model is linear and time-invariant. Through the use of the dynamic model, the future course of the controlled variables can be expressed as a function of future changes to the manipulated variables. The minimum, future control error then results from a clearly determinable sequence of future changes to the manipulated variables.

Constraints / Limitations included
While determining the solution, it is possible for constraints that affect the process such as manipulated variable limits of actuators, or other constraints, to be taken into consideration directly.

Configurable prediction horizon / Control horizon
The number of sampled values used to calculate the sum of the predicted quadratic control error can be configured with the prediction horizon. Only the first value is used to output the manipulated variable before the optimization process restarts, taking the latest measured values into consideration.

Advantages of using MPC control systems

• Easy to use in multi-variable systems. With global optimization and prioritization of the controlled variables, expensive decoupling networks are no longer necessary.
• Manipulated variable and control variable constraints are taken into account directly in the algorithm. It is also possible to weight the manipulated variable dynamic individually.
• Highly suitable for sections with a high degree of dead time (significantly greater than system time constants) or for processes with a step response that is initially in the “wrong” direction (“inverse response” behavior)
• No separate design is necessary with disturbance variable feed forward as this is taken into account directly in the algorithm. With PID controllers, this is much more complicated.
• There are business advantages from improved product quality or increased throughput resulting from the controller’s predictions or a mode of operation which can be more closely aligned to predefined control limits.

APROL model predictive controller (MPC) performance list

• Very short response times
  Very short response times between the process variables and controller since the controller’s algorithms are operated directly on the real-time hardware (controller). The process variable is updated with actual measurements in the millisecond range.
• Time can be saved during the commissioning process through standardized faceplates, process graphics and hardware combinations.
• Simple configuration of the controller through selectable standard process dynamics
• There is the possibility of complete hardware-i-theloop simulation (HIL) on the controller/master computer level using MATLAB Simulink code generation.
• Possible to test various controller configurations or pre-configurations in simulation
• Possible to train operating personnel on the model
• SISO and MIMO control configurable as needed
• Control of single-input single-output (SISO) systems as well as multiple-input multiple-output (MIMO) systems with 10 controlled variable, 10 manipulated variable and 10 disturbance variables.
• Suitable as standard either for stable processes or marginally stable processes with integrating behavior (e.g. fill level control).
• With MIMO systems, there is also the possibility to combine characteristics (stable, critically stable or proportional, integral) in one controller configuration depending on the controlled variable.
• Integrated disturbance variable feed forward:
• Up to 10 measurable disturbance inputs configurable directly in the controller.
• If a disturbance variable process is known, future disturbances can be taken into account directly in the algorithm (data from the production/maintenance planning such as withdrawal plans).
• Possible to configure constraints for absolute manipulated variables and manipulated variable changes as well as soft constraints of the controlled variables as needed
• Structural changes possible during routine normal operation (update of process dynamics and reconfiguration)

MPC functional principle
The following information assumes a system with one controlled variable, one manipulated variable and one disturbance input. The open loop controlled variable, calculated according to the model, results from superimposing the effect of the manipulated variable on the controlled variable onto the effect of the disturbance variable on the controlled variable.

Model / System deviations or unknown disturbances (at undefined places in the process) will cause the calculated controlled variable of the open loop to deviate from the measurement. This difference is taken into consideration by introducing an additive disturbance input to the model.

Optimizing and calculating the manipulated variables
The mathematical prediction of the future course of the controlled variable is a function of how the input variables have behaved in the system in the past as well as the future manipulated variables y. Our goal is to keep the control error as minimal as possible while still honoring all constraints.

Only the first value of the optimum manipulated variable course is used
The optimal solution delivers a sequence of future changes to the manipulated variable. For the control loop, however, only the current, first part of the optimal solution is used as the manipulated variable.

 

Various pre-prepared models are available to the user to describe dynamic models. Each block has its own faceplate that enables the user to use each respective dynamic block efficiently.

1. Differentiator (DT1)
This block implements the DT1 transfer function with gain factor Gain (K) and time constant

Time-Constant (T1)
2. First-order proportional component (PT1) This block implements the PT1 transfer function with gain factor Gain (K) and time constant Time-Constant (T1).

3. Lead lag component
The lead lag component consists of one DT1 component and one PT1 component. It therefore contains a parallel connection of the 3 characteristic types P, I and D.

4. Standard transfer function
The standard transfer function combines different transfer functions from the process automation. The nature of the transfer function can be selected in a drop-down list and configured.

5. Configuring valve characteristics
The implemented functions enable the valve characteristics to be configured or calibrated individually.

The following filters are available to the user to suppress noise, dampen resonant frequencies in a universal filter block that can be configured online, etc.

1. Low pass (LP)
A low-pass filter passes signals below the cutoff frequency without dampening them.

2. High-pass filter
A high-pass filter passes signals above the cutoff frequency undamped but dampens those below it.

3. Notch
A notch filter is similar to a first-order band-stop filter. It allows signals outside of a frequency band to pass and blocks frequencies that are within it.

4. BiQuad
The biquad filter has the designation of its transfer function which consists of two quadratic polynomials, one in the numerator and one in the denominator. It is typically used as a compensation controller on controlled systems. One signal frequency can thus be amplified and the other attenuated.

5. Band-pass
A band-pass filter passes signals within a frequency band undamped but dampens those outside of it.

6. Band-stop filter
A band-stop filter is used to attenuate a specific frequency band. In contrast to the notch filter, the frequency band to be attenuated has a higher bandwidth. A bandstop filter allows signals outside of a frequency band to pass through unaltered and blocks signals that are within it.

7. Moving average
This function block generates the constantly updated average value of its input from the values last in the view window and outputs it to the Out output.

It is possible to base piecemeal linear functions on one- or two-dimensional base interpolation if needed.

 

Lookup 1D

• This block outputs the input signal x to the y output after performing linear interpolation ("f(x)" or "f-1(x)").

Lookup 2D

• This block outputs the input signal lnX and lnY to the Out output after performing linear interpolation "f(x.y)".

 

Manual mode ensures stable operation
Two-thirds of control loops are either poorly configured or operated in “Manual” mode either temporarily or permanently in order to ensure a stable control loop. Plant operators are not in a position to continuously monitor their assigned control loops, however. This can only be achieved with integrated process control system functionality. For this reason, APROL APC provides control performance monitoring for control loops.

Creeping degradation
Monitoring control quality helps detect creeping degradations of control loop performance so that maintenance measures can be taken or control parameters optimized promptly.

Statistical data identifies tendencies
The CPM01 control module in APROL APC provides various statistics that can be used to assess thequality of control loops. These figures are the basis for achieving the maximum possible increase in process control efficiency. APC solutions also depend essentially on the quality of the underlying basic automation. The basic idea is to prepare or present control loop relevant data. The identified statistics must be interpreted by the user. The minimum variance index is thus determined by comparing the variance of the controlled variable of the controller being used with the variance which would occur through the use of a minimum variance controller.

If there is noticeable drift in this MV index over a specific time period, this is an indicator that there has been a change to the controlled variable. The absolute value is not critical for the evaluation; the relative value or trend is relevant here.

The continuously calculated statistics for the CPM01 module include integral absolute error, integral squared error, standard deviation of the setpoint and control error, average value of the manipulated variable, standard deviation of the manipulated variable and the controlled variable, controlled variable setpoint crossings numerator and display of the process limit (manipulated variable saturation).

Statistics calculated according to the state include the time slice during which the controller is operated in automatic mode (ServiceFactor), the system’s dead time, the minimum variance index and control performance indices (differences between optimal and current evaluation criteria), etc.